LSTM-based Prediction of Multi-function Radar Work Mode

Multi-function radar (MFR) is widely adopted to acquire target information due to its advantage on fast agility abilities of waveform and beam in electronic countermeasure environment. It is important for predicting the MFR behavior to better enhance the efficiency of electronic interference and gain the initiative of electronic warfare. This paper therefore addresses the problem of MFR work mode prediction. Specifically, a hierarchical model revealing the relationship between MFR signals, task sequences and work modes, is first established. Then, since the lengths and types of task sequences corresponding to different work modes are different, the long short-term memory (LSTM) network is developed to learn the characteristics of these task sequences, enabling the reconnaissance party to predict the MFR work mode in a short period of time. The numerical example is designed to evaluate the performance of the proposed framework and highlights that it has a high prediction accuracy for MFR work mode.